HFT is often bandied about in the media as a greedy activity that produces no real economic value. It's an easy to believe argument - certainly no ipods are directly created and no minerals are mined. The argument doesn't seem like a proof though.
Is HFT creating economic value via creating liquidity or some other value or is it truly just theft and why?
This is a very important question and you will find arguments from both sides, in part because it is still understudied.
Ben Golub, Economics Ph.D., from Stanford answers "Is high-frequency trading good for the economy?" on Quoram quite well.
This is an important but understudied question. There are few published academic studies on it, though several groups are working on the subject. You may be interested in the following papers:
http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1569067
http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1361184
These document some of the phenomena that arise in high frequency trading, from a theoretical and an empirical perspective. However, a full equilibrium analysis of the unique features of high frequency trading is still missing, and until it is done, all our answers will be kind of tentative.
Nevertheless, there are some obvious things one can say. Currently, high frequency traders are competing to locate physically closer and closer to exchanges, because milliseconds matter. Thus, large amounts of money are being spent to beat other market makers by tiny fractions of a second. Once many firms make these investments, the market looks like it did before in terms of competition and prices, but is a tiny bit faster. This investment is unlikely to be socially efficient: that is, the users of the market don't actually benefit from the fact that their trades are executed half a millisecond faster -- certainly not enough to cover all the investment that went into making that happen.
Some people who study the issue believe that high frequency trading (HFT) actually exacerbates market volatility; some plots to this effect are found in the second paper linked above. There is certainly no widely accepted theory that says faster trading technology necessarily increases efficiency, and it is easy to think of algorithms that can make money (at least in the short run) but hurt most other investors, as well as the informational value of the market.
One caution is that some of the complaining about HFT comes from those who lose when HFT gets better -- old-style market makers. They certainly have an incentive to make HFT out to be very bad. So some complaints about the predatory nature of HFT should be taken with a grain of salt. There is no strong economic consensus about the value of this activity. For what it's worth, my personal impression is that this is more bad than good. I'll post an update here as more definitive research comes out.
You can also find a debate on High-frequency trading from the Economist which gives both sides of the argument.
In conclusion: Regardless of how you feel about HFT it seems like it's here to stay and won't be leaving in the foreseeable future. So the debate will rage on...
Additional resource you may finding interesting:
You pointed out that HFT does not create ipods are mine minerals. Neither does human trading.
HFT is a proxy for human trading. Although the computer is executing trades automatically based on an algorithm, it is still using money from a human being's account so the trading is still being done with someone's money.
Fast execution of trades is desirable in exchanges. Imagine two exchanges: One only executes trades once a month, the other executes trades once a week. Which exchange would be more desirable? The exchange that trades once a week. Why? Because if I'm holding a stock that I would like to sell, I want to sell it now - not a month from now. Same reason for buying. This concept works all the way down to seconds and fractions of seconds.
The issue with HFT, however, is there are cases where the market goes against the HFT algorithm and the algorithm continues to execute trades driving prices up or down by large amounts in the matter of minutes or even seconds. The exchange frequently cancels these trades which only encourages more aggressive HFT trading since HFT traders can have their losses cancelled. This is a privilege that LFTs (low frequency traders) do not receive. This is a valid criticism of HFTs.
A short list of such cancelled trades:
8/26/2010: Nasdaq cancels trades of CORE stock
10/4/2010: Nasdaq cancels trades of CENX stock
10/15/2010: NYSE cancels trades of PAY stock
10/18/2010: NYSE cancels $500 million worth of SPY trades
5/18/2011: NYSE cancels 15,900 trades of BEE.PR.C
6/21/2011: Nasdaq cancels CNTY trades
12/2/2011: London Metals Exchange cancel trades of copper
